深騰7000上用bsub命令提交MPI作業

來源:互聯網
上載者:User

小規模作業運行在厚節點隊列上,配置:

這三個隊列節點都一樣,16路4核 共64核 Xeon X7350 2.93GHz, 512G記憶體
x64_small 共2個節點 1-8核 6小時
x64_3950 共5個節點 1-64核 6小時
x64_3950_long 共11個節點 1-64核 144小時 

x64_small就是用來給小作業啟動並執行。
如果需要運行稍大作業的話,不妨用x64_3950或x64_3950_long,事實上這兩個隊列的資源使用方式相對x64_small更為空白閑一 些

中大規模作業運行在刀片隊列,限制核心數>= 64


作業已耗用時間1分鐘,需要2個CPU核心,單個節點上使用1個CPU核心,提交到x64_small 隊列,標準輸出檔案為zlt.out,錯誤輸出檔案為zlt.err,運行程式名為comm:

[scwangj@LB270108 zjl]$ bsub -W 1 -a intemmpi -n 2 -R "span[ptile=1]" -q x64_small -o zlt.out -e zlt.err mpirun.lsf ./comm
Job <78607> is submitted to queue <x64_small>.

一次提交多個作業,寫個bash指令碼submit.sh[其實不必這樣,命令疊加的方式也不錯]:

#!/bin/bashfor i in 50 60 70 80 90 100do    bsub -W 6 -a intemmpi -n $i -R span[ptile=1] -q x64_blades -o $i.out -e $i.err ./matrixdone

查看作業:

[scwangj@LB270210 zl]$ bjobs -u scwangjJOBID    USER    STAT  QUEUE      FROM_HOST   EXEC_HOST   JOB_NAME   SUBMIT_TIME82504    scwangj PEND  x64_blades lb270210                ./matrix   Jul  4 12:4782505    scwangj PEND  x64_blades lb270210                ./matrix   Jul  4 12:4782506    scwangj PEND  x64_blades lb270210                ./matrix   Jul  4 12:4782507    scwangj PEND  x64_blades lb270210                ./matrix   Jul  4 12:4782487    scwangj PEND  x64_small  lb270210                ./matrix   Jul  4 12:35[scwangj@LB270210 zl]$ 

附:

[scwangj@v3903 20x20x100]$ cat submit.sh #!/bin/bashfor i in 1 2 3 4 5 6 7 8do    bsub -W 5:40 -a intelmpi -n $i -R span[ptile=2] -q x64_small -o $i.out -e $i.err mpirun.lsf ./simpledone[scwangj@v3903 20x20x100]$ cd ..[scwangj@v3903 ddm]$ ls10.err  18.err  1.err  20x20x100  2.out  9.out  bsubmpi   ddm.sh  fluid.grd   serial  solveuss.F  solvewss.F  stagsimple.F  submit.log  tdma.F    uc.fun  variable.mod10.out  18.out  1.out  2.err      9.err  a.sh   bsub.txt  del.sh  ppoisson.F  simple  solvevss.F  s.sh        submit2.sh    submit.sh   time.dat  uc.nam[scwangj@v3903 ddm]$ cat submit.sh #!/bin/bashfor i in 1 2 3 4 5 6 7 8do    bsub -W 5:40 -a intelmpi -n $i -R span[ptile=2] -q x64_small -o $i.out -e $i.err mpirun.lsf ./simpledone[scwangj@v3903 ddm]$ cat submit2.sh #!/bin/bashfor i in  9 10 11 12 13 14 15 16 17 18do    bsub -W 5:40 -a intelmpi -n $i -R span[ptile=9] -q x64_3950 -o $i.out -e $i.err mpirun.lsf ./simpledone[scwangj@v3903 ddm]$ bjobs -u scwangjJOBID    USER    STAT  QUEUE      FROM_HOST   EXEC_HOST   JOB_NAME   SUBMIT_TIME82725    scwangj RUN   x64_3950   v3903       9*t3701     * ./simple Jul  4 20:39                                              2*t360182726    scwangj RUN   x64_3950   v3903       9*t3802     * ./simple Jul  4 20:39                                              3*t410282727    scwangj RUN   x64_3950   v3903       9*t3701     * ./simple Jul  4 20:39                                              4*t380282728    scwangj RUN   x64_3950   v3903       9*t3601     * ./simple Jul  4 20:39                                              5*t410282729    scwangj RUN   x64_3950   v3903       9*t3701     * ./simple Jul  4 20:39                                              6*t380282730    scwangj RUN   x64_3950   v3903       9*t3601     * ./simple Jul  4 20:39                                              7*t410282731    scwangj RUN   x64_3950   v3903       9*t3701     * ./simple Jul  4 20:39                                              8*t380282745    scwangj RUN   x64_3950   v3903       9*t3701     * ./simple Jul  4 20:4182634    scwangj RUN   x64_small  v3903       t4601       * ./simple Jul  4 16:5982635    scwangj RUN   x64_small  v3903       1*t4601     * ./simple Jul  4 16:59                                              1*t370182710    scwangj RUN   x64_small  v3903       t4601       * ./simple Jul  4 20:3782711    scwangj RUN   x64_small  v3903       2*t3701     * ./simple Jul  4 20:3782746    scwangj PEND  x64_3950   v3903                   * ./simple Jul  4 20:4182619    scwangj PEND  x64_small  lb270210                * ./matrix Jul  4 16:53[scwangj@v3903 ddm]$ 

聯繫我們

該頁面正文內容均來源於網絡整理,並不代表阿里雲官方的觀點,該頁面所提到的產品和服務也與阿里云無關,如果該頁面內容對您造成了困擾,歡迎寫郵件給我們,收到郵件我們將在5個工作日內處理。

如果您發現本社區中有涉嫌抄襲的內容,歡迎發送郵件至: info-contact@alibabacloud.com 進行舉報並提供相關證據,工作人員會在 5 個工作天內聯絡您,一經查實,本站將立刻刪除涉嫌侵權內容。

A Free Trial That Lets You Build Big!

Start building with 50+ products and up to 12 months usage for Elastic Compute Service

  • Sales Support

    1 on 1 presale consultation

  • After-Sales Support

    24/7 Technical Support 6 Free Tickets per Quarter Faster Response

  • Alibaba Cloud offers highly flexible support services tailored to meet your exact needs.